CN109859228A - Recommend the method, apparatus and computer equipment of shoes money type based on image recognition - Google Patents
Recommend the method, apparatus and computer equipment of shoes money type based on image recognition Download PDFInfo
- Publication number
- CN109859228A CN109859228A CN201910047034.9A CN201910047034A CN109859228A CN 109859228 A CN109859228 A CN 109859228A CN 201910047034 A CN201910047034 A CN 201910047034A CN 109859228 A CN109859228 A CN 109859228A
- Authority
- CN
- China
- Prior art keywords
- toe
- user
- foot
- image
- type
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
Abstract
The present invention is suitable for field of artificial intelligence, provides a kind of method, apparatus and computer equipment for recommending shoes money type based on image recognition.Wherein, method includes: at least original image for obtaining user foot；Gray proces are carried out to original image, obtain gray level image；Edge processing is carried out to gray level image, extracts marginal value, the toe characteristic information of user is obtained according to marginal value；Toe characteristic information and the sample toe characteristic information in default first database are compared, similarity value is obtained；When similarity value is more than reservation threshold, then matched sample corresponding with toe characteristic information is marked, determines that the foot type of user foot is the foot type of matched sample, export foot type；According to the foot type of user, shoes money type corresponding with foot type in default second database is called；Shoes money type is recommended into user.The present invention can export the accuracy that recommendation is improved to user according to the corresponding shoes money type of foot's photo Rapid matching of user.
Description
Technical field
The invention belongs to field of artificial intelligence more particularly to a kind of sides for recommending shoes money type based on image recognition
Method, device, storage medium and computer equipment.
Background technique
Clothing, food, lodging and transportion -- basic necessities of life be live with people it is closely bound up, wherein wear the comfortable shoes of a pair of in different scenes
The user of middle work has biggish meaning.Due to the quickening of modern life rhythm, user is typically chosen in online purchase shoes
Son, and be usually that shoes are selected according to the length of oneself foot, alternatively, further being adjusted according to the girth of a garment degree of foot
The size of whole shoes.
In conclusion user, when shoes are selected in online shopping, recommender system is usually according to the length of foot, width and sole
Thickness recommends shoes to user, for selection by the user.However, often will appear after selecting shoes according to above method
Selected shoes do not fit, and reduce user experience.
Summary of the invention
In view of this, the embodiment of the invention provides recommend the method, apparatus of shoes money type, storage to be situated between based on image recognition
Matter and computer equipment handle low efficiency to solve the problems, such as that business handling method exists in the prior art.
The first aspect of the embodiment of the present invention provides a kind of method for recommending shoes money type based on image recognition, comprising:
Obtain an at least original image for user foot；
Gray proces are carried out to the original image, obtain gray level image；
Edge processing is carried out to the gray level image, extracts marginal value, is believed according to the toe feature that marginal value obtains user
Breath；
The toe characteristic information and the sample toe characteristic information in default first database are compared, phase is obtained
Like angle value；
When similarity value is more than reservation threshold, then matched sample corresponding with the toe characteristic information is marked, determined
The foot type of user foot is the foot type of the matched sample, exports the foot type；
According to the foot type of user, shoes money type corresponding with the foot type in default second database is called；
The shoes money type is recommended into user.
The second aspect of the embodiment of the present invention provides a kind of device for recommending shoes money type based on image recognition, comprising:
Module is obtained, for obtaining an at least original image for user foot；
Gradation processing module obtains gray level image for carrying out gray proces to the original image；
Edge processing module is extracted marginal value, is obtained according to marginal value for carrying out edge processing to the gray level image
The toe characteristic information of user；
Contrast module, for by the sample toe characteristic information in the toe characteristic information and default first database into
Row comparison, obtains similarity value；
First mark module, for when similarity value is more than reservation threshold, then marking and the toe characteristic information pair
The matched sample answered determines that the foot type of user foot is the foot type of the matched sample, exports the foot type；
First calling module calls corresponding with the foot type in default second database for the foot type according to user
Shoes money type；
Recommending module, for the shoes money type to be recommended user.
The third aspect of the embodiment of the present invention provides a kind of computer equipment, comprising: memory, processor and storage
In the memory and the computer program that can run on the processor, the processor execute the computer program
When perform the steps of
Obtain an at least original image for user foot；
Gray proces are carried out to the original image, obtain gray level image；
Edge processing is carried out to the gray level image, extracts marginal value, is believed according to the toe feature that marginal value obtains user
Breath；
The toe characteristic information and the sample toe characteristic information in default first database are compared, phase is obtained
Like angle value；
When similarity value is more than reservation threshold, then matched sample corresponding with the toe characteristic information is marked, determined
The foot type of user foot is the foot type of the matched sample, exports the foot type；
According to the foot type of user, shoes money type corresponding with the foot type in default second database is called；
The shoes money type is recommended into user.
The fourth aspect of the embodiment of the present invention provides a kind of computer readable storage medium, the computer-readable storage
Media storage has computer program, and the computer program performs the steps of when being executed by processor
Obtain an at least original image for user foot；
Gray proces are carried out to the original image, obtain gray level image；
Edge processing is carried out to the gray level image, extracts marginal value, is believed according to the toe feature that marginal value obtains user
Breath；
The toe characteristic information and the sample toe characteristic information in default first database are compared, phase is obtained
Like angle value；
When similarity value is more than reservation threshold, then matched sample corresponding with the toe characteristic information is marked, determined
The foot type of user foot is the foot type of the matched sample, exports the foot type；
According to the foot type of user, shoes money type corresponding with the foot type in default second database is called；
The shoes money type is recommended into user.
The present invention is carried out gray proces to the original image and is obtained by an at least original image for acquisition user foot
To gray level image；Edge processing is carried out to the gray level image, extracts marginal value, characteristic value moment of distribution is calculated according to marginal value
Battle array；The characteristic value distribution matrix of the characteristic value distribution matrix and the sample in default first database is compared, when with
When the similarity of a certain sample in the first database is more than reservation threshold, it is determined that user's foot type is corresponding for the sample
Foot type, export the foot type, according to the foot type of user, call shoes money class corresponding with the foot type in default second database
The shoes money type is recommended user by type.As it can be seen that the present invention can be corresponding according to foot's photo Rapid matching of user
Shoes money type export the accuracy that recommendation is improved to user, solve that shoes money type in the prior art is improper to be caused to use
Experience low problem in family.
Detailed description of the invention
It to describe the technical solutions in the embodiments of the present invention more clearly, below will be to embodiment or description of the prior art
Needed in attached drawing be briefly described, it should be apparent that, the accompanying drawings in the following description is only of the invention some
Embodiment for those of ordinary skill in the art without any creative labor, can also be according to these
Attached drawing obtains other attached drawings.
Fig. 1 is the application environment schematic diagram for recommending shoes money type method in the embodiment of the present invention according to feet shape；
Fig. 2 is the implementation process signal for recommending shoes money type method according to feet shape that the embodiment of the present invention one provides
Figure；
Fig. 3 is the implementation process signal for recommending shoes money type method according to feet shape that one embodiment of the invention provides
Figure；
Fig. 4 be another embodiment of the present invention provides according to feet shape recommend shoes money type method implementation process signal
Figure；
Fig. 5 is a hardware structural diagram of machine learning model training device of the embodiment of the present invention；
Fig. 6 is the schematic diagram provided by Embodiment 2 of the present invention for recommending shoes money types of devices according to feet shape；
Fig. 7 is one embodiment of the invention offer according to edge processing module in feet shape recommendation shoes money types of devices
Schematic diagram；
Fig. 8 be another embodiment of the present invention provides according to feet shape recommend shoes money types of devices in second acquisition unit
Schematic diagram；
Fig. 9 is the schematic diagram for the computer equipment that the embodiment of the present invention three provides.
Specific embodiment
In being described below, for illustration and not for limitation, the tool of such as particular system structure, technology etc is proposed
Body details, to understand thoroughly the embodiment of the present invention.However, it will be clear to one skilled in the art that there is no these specific
The present invention also may be implemented in the other embodiments of details.In other situations, it omits to well-known system, device, electricity
The detailed description of road and method, in case unnecessary details interferes description of the invention.
In order to illustrate technical solutions according to the invention, the following is a description of specific embodiments.
Embodiment one
The implementation process according to feet shape recommendation shoes money type method that Fig. 1 shows the offer of the embodiment of the present invention one is shown
It is intended to.As shown in Figure 1, this according to feet shape recommend shoes money type method specifically include it is as follows:
It is provided by the present application that shoes money type method is recommended according to feet shape, it can be applicable in the application environment such as Fig. 1,
In, computer equipment is communicated by network with server.Wherein, computer equipment/client can be, but not limited to various
Personal computer, laptop, smart phone, tablet computer and portable wearable device.Server can be with independent
The server cluster of server either multiple servers composition is realized.
In one embodiment, as shown in Figure 1, provide it is a kind of according to feet shape recommend shoes money type method, in this way
It applies and is illustrated for the server in Fig. 1, include the following steps:
101, an at least original image for user foot is obtained.
102, gray proces are carried out to the original image, obtains gray level image.
For 101 and 102, in the present embodiment, can be by user using mobile terminal (such as mobile phone) to foot into
Row shooting.Specifically, a general foot type region is provided on app on mobile terminals, when user is whole using the movement
When end takes pictures to foot, the mobile mobile terminal falls into foot inside the foot type region, further, mobile
Terminal can also prompt user that first foot is put on planar object to take pictures again.It, can be by foot as the another way of shooting
It places on the touchscreen, obtains the image of foot automatically in the range that touch screen contacts according to foot.As the another way of shooting,
Can also be in such a way that toe head and foot's profile combine, so more precisely, matching degree is also higher.
It should be noted that foot herein is the light foot for not wearing socks or shoes.
103, edge processing is carried out to the gray level image, extracts marginal value, toe characteristic information is obtained according to marginal value.
Wherein, toe characteristic information is characterized Distribution value matrix, and characteristic value distribution matrix includes: the length of each toe head
Length difference between adjacent toe head.
Wherein, the length difference between adjacent toe head is calculated according to the length of each toe head correspondence.
As an embodiment of the present invention, as shown in Fig. 2, carrying out edge processing to the gray level image, edge is extracted
Value obtains toe characteristic information according to marginal value, comprising:
201, edge processing is carried out to the gray level image, obtains the toe head for being smoothly connected composition of five toe heads
Curved section.
202, rectangular coordinate system is established, the curved section of the toe head is analyzed in rectangular coordinate system, obtains institute
State the number of the pole of curved section.
203, according to the toe head where the number of the pole and the corresponding abscissa of the pole, image is determined
Characteristic information, the characteristic information include the number of pole and the title of the toe head where the corresponding abscissa of pole, described
The title of toe head includes at least halluces, food toe, middle toe, unknown toe and little toe.
As an embodiment of the present invention, as shown in figure 3, by the curved section of the toe head in rectangular coordinate system into
Row is analyzed
301, two-dimensional Cartesian coordinate system is established, the length side of axis of ordinates and toe head in the two-dimensional Cartesian coordinate system
To parallel.
302, the curved section of the toe head is shown in rectangular coordinate system, to obtain toe characteristic information.
104, the toe characteristic information and the sample toe characteristic information in default first database are compared, is obtained
To similarity value.
105, when similarity value is more than reservation threshold, then corresponding matched sample is marked, determines that user's foot type is described
The foot type of matched sample exports the foot type.
Wherein, the foot type of people can be divided into according to the distribution of lengths situation of each toe in five toes Kai Erte foot, angstrom
And foot, Rome foot, Greece's foot, German foot and square-foot are these six types of, not meeting this several class is exactly the lopsided foot of comparison, and Greece's foot is suitable
Suitable narrower shoes, such as pointed shoes, and Rome foot, German foot, the Kai Erte foot also usual toe width of square foot are larger,
Uncomfortable suitable pointed shoes.
106, according to the foot type of user, shoes money type corresponding with the foot type in default second database is called.
107, the shoes money type is recommended into user.
Before the shoes money type is recommended user, further includes:
Obtain the preference information of user；
Accordingly, it as shown in figure 4, according to the foot type of user, calls corresponding with the foot type in default second database
The shoes money type is recommended user by shoes money type, comprising:
401, according to the foot type of user, corresponding with the foot type of the user first is filtered out out of default second database
Shoes money type.
402, it according to the preference information of user, is filtered out from the first shoes money type matched with the preference information
Second shoes money type.
Wherein it is possible to by the browsing information on the shopping APP or browser of crawl user in the terminal, to institute
It states browsing information and carries out the cluster acquisition preference information.
Before step S101, further includes:
108, call the first preset function in compressed package by target image gray processing using computer vision library opencv
Processing, obtains gray level image.
Wherein, the first preset function is CvtColor, and CvtColor is the color space conversion function in Opencv, can be with
RGB color is realized to HSV, the conversion of the color spaces such as HSI can also be converted to gray level image.
109, the second preset function in the compressed package is called to handle gray level image progress marginalisation to obtain marginalisation
Image, and extract the marginal value of the marginalisation image.
Wherein, the second preset function is Laplacian operator function.
110, the pixel that all pixels value in the marginalisation image is higher than preset threshold is extracted using recursive algorithm.
111, based on the feature Distribution value of the pixel, the eigenmatrix of the feature Distribution value is extracted.
112, the image having is marked to be determined as training sample by described.
113, preliminary classification model is established, the preliminary classification model is trained by the training sample, is obtained
Disaggregated model after training.
114, it is the first foot type by the image tagged of this kind of characteristic value in disaggregated model after training, obtains every kind first
The similarity threshold values of foot type.
For 108,109 and 110, specifically, comprising: edge extracting, this reality are carried out to the gray level image after removal noise
Applying example also Laplacian operator function can be used to carry out edge extracting to image.Convolution is carried out to image using 3 × 3 template,
Its zero crossing is found, as its marginal point.Exist since the edge of extraction has breakpoint it is necessary to being connected together originally due to side
Edge extract and generate breakpoint the case where repaired.Need the carry out marginal growth method to the place of breakpoint, the side of growth
To some neighborhood of the deflection of the line segment at the place along the breakpoint, if occurring a plurality of candidate in some direction of growth
Line segment just uses priority algorithm, considers the length and deflection of candidate line sections, selects that length is longer, deflection is close at breakpoint
Deflection as growth object.It can be obtained by complete edge image in this way.
Many features involved in field of image processing, corner feature, edge feature, shape feature, textural characteristics, color are special
Sign, histogram statistical features etc..These features are to compare the feature of bottom, such as corner feature a bit, edge feature, and color is special
Sign etc., some are then more high-rise feature, such as shape feature, textural characteristics, histogram statistical features.What the present embodiment was related to
It is the edge feature in low-level image feature, the means for extracting these features are known as Edge Gradient Feature or edge detection.Edge detection
Be divided into single order detective operators and second order detective operators in common operator, operator described herein some similar to the differential in mathematics
Concept.Another form of edge detection is also referred to as phase equalization, this concept I to referring to again below, have this
Us are helped to analyze this process of edge extracting from image frequency domain after a concept.
The specific method of edge detection includes single order edge detection and second order edge detection and other side edge detections
Method.Wherein, single order edge detection includes Roberts crossover operator, Prewitt operator, Sobel operator and Canny operator；Single order
Edge detection includes Lapacian operator, Marr-Hildreth operator and LapLacian of Gaussian operator；Other sides
Edge detection method includes Spacek operator, Petrou operator and Susan operator.Analysis based on edge detection is not vulnerable to whole light
According to the influence of Strength Changes, while being easy to highlight target information using marginal information and achieving the purpose that simplify processing, therefore very
More image understanding methods are all based on edge.Edge detection is it is emphasised that picture contrast.The understanding of contrast intuitively
It is exactly the size of difference, is exactly the difference of gray value (brightness value) for gray level image.Image can be enhanced in these differences
In boundary characteristic because these boundaries are exactly the biggish embodiment of picture contrast.Here it is us to perceive the big of object boundary
Body mechanism, because the performance of target is exactly and the difference in brightness around it.Wherein, brightness change can be by carrying out consecutive points
Difference processing enhances.Carrying out difference processing to the consecutive points of horizontal direction can detecte the brightness change in vertical direction, root
It is commonly known as horizontal edge detective operators (horizontal edge detector) according to its effect, thus can detecte out
Vertical edge；Carrying out difference processing to the consecutive points of vertical direction can detecte the brightness change in horizontal direction, according to its work
With commonly known as vertical edge detective operators (vertical edge detector), it thus can detecte out horizontal sides
Edge.Horizontal edge detective operators and vertical edge detective operators are combined, so that it may while detecting vertical edge and horizontal sides
Edge.
It is understood that using Taylor series analysis it is known that the difference of adjacent two o'clock is the estimation of first derivative
Value.If being inserted into a pixel between two neighboring difference point to realize, it is equivalent to and is made with the first-order difference of two consecutive points
For new differential horizontal, it is known that the estimated value of first differential is two separated by a pixel using Taylor series analysis
The difference of point.The basis that Roberts crossover operator is realized is single order edge detection, using two templates, calculate on diagonal line and
It is not the differential of two pixels in reference axis.
Wherein, for Prewitt edge detection operator edge detection similar to differential process, the part for the variation that it is detected is inevitable
There is respective handling to the brightness change of noise and image.Therefore, average value processing is added in edge detection process and is had to
Very with caution.For example, vertical formwork Mx can be extended to three rows, and horizontal shuttering My is extended to three column.Thus obtain
Prewitt edge detection operator.Further, if the weight of two Prewitt template operator center pixels is gone twice
Numerical value, just obtain Sobel edge detection operator, it determines that two masks at edge form by approach vector.Sobel ratio
Other edge detection operator performances of the same period such as Prewitt operator are more preferable.
Wherein, the common version of Sobel operator combines on optimal smoothing and another reference axis in a reference axis
Optimal difference.It should be noted that the benefit of big edge detection template be it reduce noise smooth effect it is more preferable.
Wherein, Canny edge detection operator is formed by three main targets: the optimal detection without additional response detects side
Distance is the smallest between edge position and actual edge position is properly positioned, and reduces the multiple response of single edges and obtains single response,
Gauss operator is optimal to picture smooth treatment.The step of Canny edge detection is generally handled can be divided into following four step
It is rapid: to apply Gaussian smoothing, using Sobel operator, (non-maxima suppression is substantially to find side using non-maxima suppression
Highest point in edge intensity data), to connect marginal point, (threshold process needs two threshold values, i.e. upper limit threshold for hysteresis threshold processing
Value and lower threshold).
Wherein, the premise of single order edge detection is that differential process can be such that variation enhances.Look for image change rate maximumly
Side can not only be found by the extreme value of single order change rate, while can also be found by the zero crossing that second order changes.Second order
The difference that differential can use two adjacent first differentials is come approximate.This is also consistent with the concept in mathematics.If level
Second-order Operator and disposition Second Order Differential Operator combine, available one full Laplacian template operator.Marr-
Hidreth is also with gaussian filtering.The surface chart of the operator is the shape of mexican hat, so being also known as " ink sometimes
Western brother's cap " operator, in fact, if Gaussian smoothing and Laplacian operator are combined, an available LoG
(Laplacian of Gaussian) operator, it is exactly the base of Marr-Hidreth.
For 111,112,113 and 114, wherein Fig. 5 illustrates one of machine learning model training device 10
Optional hardware structural diagram, comprising: processor 11, input/output interface 13, storage medium 14 and network interface 12,
Component can be through 15 connection communication of system bus.Processor 11 can using central processing unit (CPU), microprocessor (MCU,
Microcontroller Unit), specific integrated circuit (ASIC, Application Specific Integrated
Circuit) or logic programmable gate array (FPGA, Field-Programmable Gate Array) is realized.Input/output
Interface 13 can be realized using such as display screen, touch screen, loudspeaker input/output device.Storage medium 14 can be using sudden strain of a muscle
Deposit, the non-volatile memory mediums such as hard disk, CD are realized, can also be using dual-magnification technique (DDR, Double Data Rate) dynamic
The volatile storage mediums such as caching realize that illustratively, storage medium 14 can be common with machine learning model training device 10
It is arranged in same place, can also be arranged relative to 10 place remote of machine learning model training device, or opposite engineering
Practise the distribution setting of 10 local and remote side of model training apparatus.Network interface 12 provides external data such as strange land to processor 11 and is arranged
Storage medium 14 access ability, illustratively, network interface 12 can be based on near-field communication (NFC, Near Field
Communication) the short-range communication that technology, bluetooth (Bluetooth) technology, purple honeybee (ZigBee) technology carry out, in addition,
It can also realize as based on CDMA (CDMA, Code Division Multiple Access), wideband code division multiple access
The communication of communication standards such as (WCDMA, Wideband Code Division Multiple Access) and its evolution standard.
This is proposed based on the hardware configuration of above-mentioned machine learning model training method and machine learning model training device
Inventive embodiments.
It should be understood that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment, each process
Execution sequence should be determined by its function and internal logic, the implementation process without coping with the embodiment of the present invention constitutes any limit
It is fixed.
Embodiment two
Referring to FIG. 6, it illustrates the devices provided by Embodiment 2 of the present invention for recommending shoes money type based on image recognition
Schematic diagram.The device 60 for recommending shoes money type based on image recognition, comprising: acquisition module 61, gradation processing module 62,
Edge processing module 66, contrast module 64, the first mark module 65, the first calling module 66 and recommending module 67.Wherein, each mould
The concrete function of block is as follows:
Module 61 is obtained, for obtaining an at least original image for user foot.
Gradation processing module 62 obtains gray level image for carrying out gray proces to the original image.
Edge processing module 63 is extracted marginal value, is obtained according to marginal value for carrying out edge processing to the gray level image
Take the toe characteristic information at family.
Contrast module 64, for by the sample toe characteristic information in the toe characteristic information and default first database
It compares, obtains similarity value.
First mark module 65, for when similarity value is more than reservation threshold, then marking and the toe characteristic information
Corresponding matched sample determines that the foot type of user foot is the foot type of the matched sample, exports the foot type.
First calling module 66 calls corresponding with the foot type in default second database for the foot type according to user
Shoes money type.
Recommending module 67, for the shoes money type to be recommended user.
Optionally, as shown in fig. 7, edge processing module 63 includes:
First acquisition unit 630 obtains smoothly connecting for five toe heads for carrying out edge processing to the gray level image
Connect the curved section of the toe head of composition；
Second acquisition unit 632 obtains institute for analyzing the curved section of the toe head in rectangular coordinate system
State the number of the pole of curved section；
Determination unit 634, for the toe where the number and the corresponding abscissa of the pole according to the pole
Head determines the characteristic information of image, the number and the toe where the corresponding abscissa of pole that the characteristic information includes pole
The title of head, the title of the toe head include at least halluces, food toe, middle toe, unknown toe and little toe.
Optionally, as shown in figure 8, second acquisition unit 632 includes:
Establish unit 6321, for establishing two-dimensional Cartesian coordinate system, axis of ordinates in the two-dimensional Cartesian coordinate system with
The length direction of toe head is parallel；
Display unit 6322, for showing the curved section of the toe head in rectangular coordinate system.
Optionally, the first calling module 66 includes:
First screening unit 661 filters out and the user out of default second database for the foot type according to user
The corresponding first shoes money type of foot type；
Second screening unit 662 filters out and institute for the preference information according to user from the first shoes money type
State the matched second shoes money type of preference information；
Wherein, the browsing information done shopping on APP or browser by crawl user in the terminal, to described clear
Information of looking at carries out cluster and obtains the preference information.
Optionally, recommend the device of shoes money type based on image recognition further include:
Second calling module, for calling the first preset function in compressed package by mesh using computer vision library opencv
The processing of logo image gray processing, obtains gray level image；
First extraction module, for calling the second preset function in the compressed package to carry out gray level image at marginalisation
Reason obtains marginalisation image, and extracts the marginal value of the marginalisation image；
Second extraction module, for using recursive algorithm extract in the marginalisation image all pixels value be higher than it is default
The pixel of threshold value；
Third extraction module extracts the spy of the feature Distribution value for the feature Distribution value based on the pixel
Levy matrix；
Determining module, for marking the image having to be determined as training sample for described；
Module is established, for establishing preliminary classification model, the preliminary classification model is carried out by the training sample
Training, the disaggregated model after being trained；
Second mark module, in disaggregated model after training by the image tagged of this kind of characteristic value be the first foot
Type obtains the similarity threshold values of every kind of first foot type.
The device provided in an embodiment of the present invention for recommending shoes money type based on image recognition, by obtaining user foot extremely
A few original image carries out gray proces to the original image and obtains gray level image；Edge is carried out to the gray level image
Marginal value is extracted in processing, calculates characteristic value distribution matrix according to marginal value；By the characteristic value distribution matrix and default first number
It is compared according to the characteristic value distribution matrix of the sample in library, when super with the similarity of a certain sample in the first database
When crossing reservation threshold, it is determined that user's foot type is the corresponding foot type of the sample, exports the foot type, according to the foot type of user,
Shoes money type corresponding with the foot type in default second database is called, the shoes money type is recommended into user.As it can be seen that this
Invention can export the standard that recommendation is improved to user according to the corresponding shoes money type of foot's photo Rapid matching of user
Exactness solves the problems, such as that shoes money type is improper in the prior art and causes user experience low.
Embodiment three
Fig. 9 is a kind of computer equipment that the embodiment of the present invention three provides, including memory, processor and is stored in storage
On device and the computer program that can run on a processor, processor realize basis in above-described embodiment when executing computer program
Feet shape recommends the step of shoes money type method, such as step 101 shown in Fig. 2 to step 107.Alternatively, processor executes
The function of recommending each module/unit of shoes money types of devices in above-described embodiment according to feet shape is realized when computer program,
Such as module 61 shown in Fig. 6 is to the function of module 67.To avoid repeating, which is not described herein again.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program realizes the step of recommending the method for shoes money type based on image recognition in above-described embodiment when being executed by processor, such as
Step 101 shown in Fig. 2 is to step 105.Alternatively, being realized when computer program is executed by processor in above-described embodiment based on figure
As identification recommends the function of each module/unit of the device of shoes money type, such as module 61 shown in Fig. 7 to the function of module 65.
To avoid repeating, which is not described herein again.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
It is apparent to those skilled in the art that for convenience of description and succinctly, only with above-mentioned each function
Can unit, module division progress for example, in practical application, can according to need and by above-mentioned function distribution by different
Functional unit, module are completed, i.e., the internal structure of described device is divided into different functional unit or module, more than completing
The all or part of function of description.
In embodiment provided by the present invention, it should be understood that disclosed device/terminal device and method, it can be with
It realizes by another way.For example, device described above/terminal device embodiment is only schematical, for example, institute
The division of module or unit is stated, only a kind of logical function partition, there may be another division manner in actual implementation, such as
Multiple units or components can be combined or can be integrated into another system, or some features can be ignored or not executed.Separately
A bit, shown or discussed mutual coupling or direct-coupling or communication connection can be through some interfaces, device
Or the INDIRECT COUPLING or communication connection of unit, it can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
Embodiment described above is merely illustrative of the technical solution of the present invention, rather than its limitations；Although referring to aforementioned reality
Applying example, invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each
Technical solution documented by embodiment is modified or equivalent replacement of some of the technical features；And these are modified
Or replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution should all
It is included within protection scope of the present invention.
Claims (10)
1. a kind of method for recommending shoes money type based on image recognition characterized by comprising
Obtain an at least original image for user foot；
Gray proces are carried out to the original image, obtain gray level image；
Edge processing is carried out to the gray level image, extracts marginal value, the toe characteristic information of user is obtained according to marginal value；
The toe characteristic information and the sample toe characteristic information in default first database are compared, similarity is obtained
Value；
When similarity value is more than reservation threshold, then matched sample corresponding with the toe characteristic information is marked, determines user
The foot type of foot is the foot type of the matched sample, exports the foot type；
According to the foot type of user, shoes money type corresponding with the foot type in default second database is called；
The shoes money type is recommended into user.
2. the method according to claim 1 for recommending shoes money type based on image recognition, which is characterized in that the gray scale
Image carries out edge processing, extracts marginal value, obtains toe characteristic information according to marginal value, comprising:
Edge processing is carried out to the gray level image, obtains the curved section of the toe head for being smoothly connected composition of five toe heads；
The curved section of the toe head is analyzed in rectangular coordinate system, obtains the number of the pole of the curved section；
According to the toe head where the number of the pole and the corresponding abscissa of the pole, the feature letter of image is determined
Breath, the characteristic information includes the number of pole and the title of the toe head where the corresponding abscissa of pole, the toe head
Title include at least halluces, food toe, middle toe, unknown toe and little toe.
3. the method according to claim 2 for recommending shoes money type based on image recognition, which is characterized in that by the toe
The curved section of head is analyzed in rectangular coordinate system, comprising:
Two-dimensional Cartesian coordinate system is established, the axis of ordinates in the two-dimensional Cartesian coordinate system is parallel with the length direction of toe head；
The curved section of the toe head is shown in rectangular coordinate system.
4. the method according to claim 1 for recommending shoes money type based on image recognition, which is characterized in that according to user's
Foot type, the interior shoes money type corresponding with the foot type of default second database of calling include:
According to the foot type of user, the first shoes money class corresponding with the foot type of the user is filtered out out of default second database
Type；
According to the preference information of user, filtered out from the first shoes money type and the matched second shoes money of the preference information
Type, the preference information of the user are the browsings done shopping on APP or browser by crawl user in the terminal
Information carries out cluster acquisition to the browsing information.
5. the method according to any one of claims 1 to 4 for recommending shoes money type based on image recognition, which is characterized in that
It is compared by the toe characteristic information and the sample toe characteristic information in default first database, obtains similarity value
Before, further includes:
It calls the first preset function in compressed package to handle target image gray processing using computer vision library opencv, obtains
Gray level image；
It calls the second preset function in the compressed package to handle gray level image progress marginalisation to obtain marginalisation image, and mentions
Take the marginal value of the marginalisation image；
The pixel that all pixels value in the marginalisation image is higher than preset threshold is extracted using recursive algorithm；
Feature Distribution value based on the pixel extracts the eigenmatrix of the feature Distribution value；
The image having is marked to be determined as training sample by described；
Preliminary classification model is established, the preliminary classification model is trained by the training sample, after being trained
Disaggregated model；
In disaggregated model after training by the image tagged of this kind of characteristic value be the first foot type, obtain every kind of first foot type phase
Like bottom valve value.
6. a kind of device for recommending shoes money type based on image recognition characterized by comprising
Module is obtained, for obtaining an at least original image for user foot；
Gradation processing module obtains gray level image for carrying out gray proces to the original image；
Edge processing module extracts marginal value, obtains user according to marginal value for carrying out edge processing to the gray level image
Toe characteristic information；
Contrast module, for carrying out pair the sample toe characteristic information in the toe characteristic information and default first database
Than obtaining similarity value；
First mark module, for when similarity value is more than reservation threshold, then marking corresponding with the toe characteristic information
Matched sample determines that the foot type of user foot is the foot type of the matched sample, exports the foot type；
First calling module calls shoes money corresponding with the foot type in default second database for the foot type according to user
Type；
Recommending module, for the shoes money type to be recommended user.
7. the device according to claim 6 for recommending shoes money type based on image recognition, which is characterized in that the edge
Managing module includes:
First acquisition unit, for carrying out edge processing to the gray level image, obtain five toe heads is smoothly connected composition
Toe head curved section；
Second acquisition unit obtains the curve for analyzing the curved section of the toe head in rectangular coordinate system
The number of the pole of section；
Determination unit is determined for the toe head where the number and the corresponding abscissa of the pole according to the pole
The characteristic information of image, the characteristic information include the number of pole and the name of the toe head where the corresponding abscissa of pole
Claim, the title of the toe head includes at least halluces, food toe, middle toe, unknown toe and little toe.
8. the device according to claim 6 for recommending shoes money type based on image recognition, which is characterized in that described second obtains
The unit is taken to include:
Subelement is established, the axis of ordinates and toe head for establishing two-dimensional Cartesian coordinate system, in the two-dimensional Cartesian coordinate system
Length direction it is parallel；
Subelement is shown, for showing the curved section of the toe head in rectangular coordinate system.
9. a kind of computer equipment, including memory, processor and storage are in the memory and can be in the processor
The computer program of upper operation, which is characterized in that the processor realized when executing the computer program as claim 1 to
The step of any one of 5 the method.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In when the computer program is executed by processor the step of any one of such as claim 1 to 5 of realization the method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910047034.9A CN109859228A (en) | 2019-01-18 | 2019-01-18 | Recommend the method, apparatus and computer equipment of shoes money type based on image recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910047034.9A CN109859228A (en) | 2019-01-18 | 2019-01-18 | Recommend the method, apparatus and computer equipment of shoes money type based on image recognition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109859228A true CN109859228A (en) | 2019-06-07 |
Family
ID=66895187
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910047034.9A Pending CN109859228A (en) | 2019-01-18 | 2019-01-18 | Recommend the method, apparatus and computer equipment of shoes money type based on image recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109859228A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112487270A (en) * | 2019-09-12 | 2021-03-12 | 北京白帽汇科技有限公司 | Method and device for asset classification and accuracy verification based on picture identification |
| CN112699783A (en) * | 2020-12-29 | 2021-04-23 | 深圳力维智联技术有限公司 | Footprint identification method, footprint identification device and computer-readable storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140285646A1 (en) * | 2012-11-08 | 2014-09-25 | Satwinder Kahlon | Apparatus for recommendation for best fitting shoe |
| CN106127773A (en) * | 2016-06-29 | 2016-11-16 | 北京三维天下科技股份有限公司 | A kind of foot type data capture method based on picture |
| CN107704873A (en) * | 2017-09-27 | 2018-02-16 | 吴立俊 | A kind of footwear money adaption system and method based on foot type and fat and lean degree |
| CN107851328A (en) * | 2015-05-18 | 2018-03-27 | Embl零售股份有限公司 | For the method and system for recommending to be adapted to footwear |
| US20180160776A1 (en) * | 2016-12-14 | 2018-06-14 | Black Brass, Inc. | Foot measuring and sizing application |
| CN108305286A (en) * | 2018-01-25 | 2018-07-20 | 哈尔滨工业大学深圳研究生院 | Multi-view stereo vision foot type method for three-dimensional measurement, system and medium based on color coding |
| CN108805138A (en) * | 2018-04-19 | 2018-11-13 | 北京金尺云数科技有限公司 | A kind of method that mobile phone photograph calculates foot's data |
| CN109214873A (en) * | 2018-10-29 | 2019-01-15 | 黄冠文 | A kind of intelligent shoe cabinet and shoes recommended method |
-
2019
- 2019-01-18 CN CN201910047034.9A patent/CN109859228A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140285646A1 (en) * | 2012-11-08 | 2014-09-25 | Satwinder Kahlon | Apparatus for recommendation for best fitting shoe |
| CN107851328A (en) * | 2015-05-18 | 2018-03-27 | Embl零售股份有限公司 | For the method and system for recommending to be adapted to footwear |
| CN106127773A (en) * | 2016-06-29 | 2016-11-16 | 北京三维天下科技股份有限公司 | A kind of foot type data capture method based on picture |
| US20180160776A1 (en) * | 2016-12-14 | 2018-06-14 | Black Brass, Inc. | Foot measuring and sizing application |
| CN107704873A (en) * | 2017-09-27 | 2018-02-16 | 吴立俊 | A kind of footwear money adaption system and method based on foot type and fat and lean degree |
| CN108305286A (en) * | 2018-01-25 | 2018-07-20 | 哈尔滨工业大学深圳研究生院 | Multi-view stereo vision foot type method for three-dimensional measurement, system and medium based on color coding |
| CN108805138A (en) * | 2018-04-19 | 2018-11-13 | 北京金尺云数科技有限公司 | A kind of method that mobile phone photograph calculates foot's data |
| CN109214873A (en) * | 2018-10-29 | 2019-01-15 | 黄冠文 | A kind of intelligent shoe cabinet and shoes recommended method |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112487270A (en) * | 2019-09-12 | 2021-03-12 | 北京白帽汇科技有限公司 | Method and device for asset classification and accuracy verification based on picture identification |
| CN112699783A (en) * | 2020-12-29 | 2021-04-23 | 深圳力维智联技术有限公司 | Footprint identification method, footprint identification device and computer-readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liang et al. | An efficient forgery detection algorithm for object removal by exemplar-based image inpainting | |
| EP3455782B1 (en) | System and method for detecting plant diseases | |
| US9898686B2 (en) | Object re-identification using self-dissimilarity | |
| Zhang et al. | A novel graph-based optimization framework for salient object detection | |
| Lalonde et al. | Detecting ground shadows in outdoor consumer photographs | |
| CN109784281A (en) | Products Show method, apparatus and computer equipment based on face characteristic | |
| CN102460475A (en) | Histogram methods and systems for object recognition | |
| Arbelot et al. | Local texture-based color transfer and colorization | |
| JP5671928B2 (en) | Learning device, learning method, identification device, identification method, and program | |
| CN111160284A (en) | Method, system, equipment and storage medium for evaluating quality of face photo | |
| CN109348731A (en) | A kind of method and device of images match | |
| CN110705634B (en) | Heel model identification method and device and storage medium | |
| WO2018189802A1 (en) | Image processing device, image processing method, and program | |
| CN109063598A (en) | Face pore detection method, device, computer equipment and storage medium | |
| Bora et al. | A new efficient color image segmentation approach based on combination of histogram equalization with watershed algorithm | |
| CN109859228A (en) | Recommend the method, apparatus and computer equipment of shoes money type based on image recognition | |
| Kapur et al. | Mastering opencv android application programming | |
| Dwina et al. | Skin segmentation based on improved thresholding method | |
| Zhang et al. | Saliency detection by selective color features | |
| Feng et al. | Image shadow detection and removal based on region matching of intelligent computing | |
| Li et al. | Two-layer average-to-peak ratio based saliency detection | |
| CN113850748A (en) | Point cloud quality evaluation system and method | |
| Wang et al. | Confidence-driven image co-matting | |
| Xu et al. | Saliency detection via bi-directional propagation | |
| CN109033797A (en) | A kind of authority setting method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190607 |